Links to Related Sites
This section is currently under development. Below are a few starting links.
Acroname is a developer and distributor of robot systems, components, books, etc. Their Brainstem robot controller card and associated components have enabled us to interface with the Garcia, the LRV, and an earlier generation of laptop-carrying robots.
Dr. Johann Borenstein is a prolific mechanical engineer who has developed the OmniTread family of robots, including the OT4 with which we have done research described here.
University of Massachusetts at Amherst. Andrew Barto, Richard Sutton, Sridhar Mahadevan, and others have been leading research that shows the broad applicability of reinforcement learning and related ideas to difficult problems. We plan to elaborate this link with more specific topics related to biobehavioral approaches and robotics.
Sebastian Thrun and Andrew Ng are faculty of Stanford University's Computer Science Department that approach robot control using methods that relate to those used in 7G. They favor use of "direct policy search", a variation of reinforcement learning that works more like a GA than biological learning. See Ng's control system for piloting a helicopter including inverted flight, and Thrun's SUV that won the DARPA Grand Challenge.
Marco Dorigo has developed a robot training approach he calls robot shaping. The method does not directly correspond to the established techniques of shaping as applied to animals, but it is still very relevant to biologically inspired methods related to 7G.
Rodney Brooks, head of the MIT A.I. lab, has been one of the most influential researchers demonstrating that robots can exhibit intelligent behavior without a symbolic or cognitive model. His initial work focused on direct programming in a model he called a "subsumption architecture", but later work has added learning and GAs to development approaches.
